Beijing Prescribes Rules for AI in its Online Medical Services



Many AI companies are concentrating on creating systems tailored for particular areas, such as healthcare, even though China hasn't yet authorized any of these systems to be made available for purchase. For their part, regulators have been attempting to walk a narrow line between supporting the economic advantages of the technology and minimizing its potentially harmful effects, which can include everything from job losses to the dissemination of false information.

The Beijing municipal health department has begun asking the public for feedback on a series of test measures intended to control internet diagnosis and treatment. Due to the ease they offer, more and more people are becoming interested in remote medical services today. To better serve the people, local administrations are increasingly encouraging internet-based digital medical services. For instance, according to Qingdao health authorities, a citywide health platform in East China's Shandong Province includes connections to 10 districts and 3,498 medical facilities.

By October 2022, more than 2,700 internet hospitals  were opened across China, offering more than 25.9 million people access to online medical care, according to the Digital China Development Report.

As certain platforms lack price criteria or qualifications for online diagnosis and treatment, further restrictions are however required in the quickly growing online medical services sector, according to reports. Therefore, to ensure safety and integrity of medical services, Beijing is in thoughts of constraining AI’s interference in medical prescriptions on the basis of a draft rule over internet-based medical services released by the city’s health authorities.

Banning AI from Writing Medical Prescriptions

One of the significant aspects in the draft is the ban on dispensing medication before an AI-generated prescription has been generated. The draft rule emphasizes respecting the ethical standards of the medical community by prohibiting the creation of prescription inventories for commercial interests. According to the proposed rule, physicians who provide online medical diagnoses are required to have at least three years of clinical medical practice experience.

Before commencing online patient consultations, doctors are obliged to complete a registration process using their genuine names in order to avoid impersonation and assure responsibility.


Doctors are also obligated to advise patients on the importance of providing accurate identification information and to avoid pretending to be someone else in order to receive medical care. Doctors are required to immediately stop providing internet services to patients when they are deemed unfit for continuing online diagnosis and direct them to seek care at actual healthcare facilities.

The importance of keeping thorough medical records created by internet providers is also emphasized in the draft rule. The proposed standards state that these records must be kept for at least 15 years.

The diagnosis procedure must also be kept on file for at least three years, including all conversations, audio files, and video recordings. According to the guideline, in order to maintain accountability and transparency, the complete information chain regarding prescriptions, pharmacy order fulfillment, and delivery must be traceable.

The commission also requires online medical providers to make their pricing schedules available for public review. Internet-based medical services are also subject to the same regulations that already apply to practitioners' fees, such as the ban on patient referrals made for profit and the mandate that patients buy their medical supplies from specific stores.

Up to September 16, the draft regulation is available for public comment, allowing interested parties and members of the public to add their knowledge and viewpoints to the regulatory framework influencing Beijing's future of online medical services. 

The draft rule, which consists of 41 articles, establishes a thorough framework governing various participants in internet-based medical services, including hospitals, doctors, patients, and supervisory entities.

It is crucial to keep in mind that Beijing views algorithms through a variety of lenses, not just the one filtering harmful content, to comprehend the current and future of AI governance in China. The new restrictions in China aim to push the design and application of Chinese LLMs in the direction of congruence with national interests, continuing an emerging trend in the country's digital governance.

Beijing’s Point of View of Algorithms

The CAC has had to face the difficult realization that it must strike a balance between censorship and control while allowing for technical advancement. This is one of the first times it has used its authority in a field that is subject to fierce geopolitical competition compared to its earlier regulatory action in the field of AI. This put China's innovation and geopolitical ambitions in direct conflict with the CAC's security-first policy. The government believes China's AI industry to be of enormous strategic and economic importance, and it has now entered a period of great uncertainty. Since October 2022, US export controls have made it more difficult for Chinese companies to obtain the cutting-edge processors required to train AI models to generate text, particularly those GPU models sold by American chip design giant Nvidia.

As a result, the new regulations rank among the CAC's most drastically revised regulations ever. Most importantly, these restrictions now exempt all R&D operations from inspection for any non-public-facing uses, including LLMs for medical diagnosis or industrial process automation. Additionally, several standards have been relaxed. Domestic AI experts have demanded significant changes to the provisions requiring businesses to verify that the training data they used was genuine and accurate to the extent necessary. Given the numerous billions of training data utilized for large language models, providing such guarantees would be a monumental undertaking.

The Chinese central and local governments are supporting model training that embraces generative AI applications in fields important to Xi Jinping's much-touted "real economy," from accelerating the discovery of new drugs and streamlining medical communications to modernizing manufacturing. However, the CAC will continue to be wary of netizen-facing chatbots.

Beijing's legislative measures regarding generative AI do not, in general, seem to pose a risk of stifling innovation or paralyzing private IT companies. Even though the CAC's mandate calls for it to focus largely on preventing ChatGPT-like products from influencing public opinion, it was forced to back down and relax the standards for LLM developers. As the government strives to achieve a balance between control and development, China's AI governance may continue to astound onlookers.

Current Issue